Research On Target Recognition And Trajectory Planning Of Welding Robot Based On Machine Vision

Automated welding technology has become a trend in the field of welding,in current industrial applications,welding operations are generally performed by manually teaching welding robots.But this kind of operation has low efficiency and relies heavily on the teaching experience of the worker,re-teaching is required when the teaching object s location is changed.Therefore,when the welding volume is large,it will seriously affect the production efficiency of the enterprise.For the above problems,this paper studies the target recognition and trajectory planning of welding robot based on machine vision.The specific work is as follows:(1)In this paper,a camera calibration model was established.The camera was first calibrated using the traditional Zhang's calibration method.Based on this,a camera calibration method based on BP neural network was proposed.The MATLAB simulation of the method was used to verify that the method has a higher accuracy than Zhang's calibration method.(2)The principle of binocular stereo vision is introduced and the spatial point 3D reconstruction formula is obtained.Then the binocular vision is used for measurement experiments and the measurement error is analyzed.At last,a circular marker is used to perform the robot hand-eye calibration experiment.The coordinates of the obtained points are compared with coordinates of points obtained by manual teaching,which proves that this method has high accuracy.(3)a series of processing of the weld image is performed,including filtering,thresholding,morphological processing,refinement,etc.Then the Hough Transform is proposed to detect the weld in the processed image and get the mathematical expression of the welding seam path.At last a stereo matching method based on Gaussian pyramid transform combined with polar line constraints is proposed,which the left and right weld image points are stereo-matched and the three-dimensional coordinates of the weld feature points are obtained from the low-scale to the high-scale.Then it is transferred to the coordinates of the robot base coordinate system by using the hand-eye calibration and the corresponding welding path is fitted.(4)The trajectory planning of the robot is studied.Firstly,the D-H parameter model of the robot is established,the forward inverse kinematics equation is obtained,and the effective workspace of the robot is simulated.Then,the TCP point at the end of the robot welding torch is calibrated by using the least squares method.Finally,in order to solve the problem of the smoothness of the trajectory in the robot welding process,a transition arc is used to replace the sharp corner in the path and a quaternion method is proposed to interpolate the posture of the robot.It is verified by simulation that the method can ensure the smoothness of the trajectory of the robot in space continuous motion.(5)The hardware platform of robot automatic welding system based on binocular vision is built,and the overall software interface design of the system is completed using the modular design concept.Then the robot automatic welding experiment is completed by using the designed software system.Compared with the manual teaching method,it is verified that the method of this paper has better accuracy.